Table 5 The results of adding and ablating patient emotions for LLM prediction.

From: A Multimodal Depression Consultation Dataset of Speech and Text with HAMD-17 Assessments

Models

Emotion

Precision

Recall

F1

Qwen2.5-7B-Instruct

wo. emotion

0.407

0.409

0.408

w. emotion

0.410

0.412

0.411

LLaMA3.1-8B-Instruct

wo. emotion

0.422

0.424

0.423

w. emotion

0.425

0.427

0.426

  1. "wo.” and “w.” denote “without” and “with”, respectively.
Back to article page